The second generation Ryzen is set to launch late Q1 2018, according to the latest roadmap presented by AMD. AMD is also planning to launch Ryzen 3 mobile APU (Raven Ridge) in the first quarter. Business-oriented Ryzen PRO mobile APUs are also to be expected early next year.
Too many rumors not to be fake. I wouldn't expect massive improvements. anything more than 20% improvement in per-core performance is a pipe dream. More cores are possible although extremely unlikely.
Oh so 20% uplift is unrealistic, but explain the massive 52% uplift from the previous generation with a 'from scratch' architecture.
Very reasonable expectations. I agree, 6-10% IPC and 12-15% higher clocks for advertised gains of 30% in certain applications, 25% across the board.*
I agree. If all they did was make them 4.0ghz/ 4.5ghz/ 5ghz (base/ boost/ max OC) and add solid ram compatibility up to 4000mhz that would be enough for a solid refresh.
Why? AMD was able to improve IPC on Ryzen by a significant amount compared to the previous gen, and that architecture is still new. You don't think there is any headroom for IPC gains with a brand new from scratch architecture on a new smaller node?
Because it's Zen on 12nm not Zen 2.Originally Posted by Behemoth777
Why? AMD was able to improve IPC on Ryzen by a significant amount compared to the previous gen, and that architecture is still new. You don't think there is any headroom for IPC gains with a brand new from scratch architecture on a new smaller node?
Come on... History proves you wrong sir.
...and? I don't see why both things can't be true. If Intel can improve IPC along with a die shrink, so can AMD.
Make that since Haswell, if you're talking about desktop-class enthusiasts.
Yeah, you're right. Broadwell was released in 2015 though IIRC, so roughly 2 years.Originally Posted by Artikbot
Make that since Haswell, if you're talking about desktop-class enthusiasts.
Broadwell brought power efficiency improvements, Skylake not much at all, Kaby Lake made Skylake clocks possible with Broadwell IPC, and Coffee Lake has been the only notable product since Haswell (which owes much of its success to the higher clocks it allowed without chucking efficiency out the window).
Things are different in HEDT. Since SB-E we've had reasonable improvements generation after generation.
While Anderson's responsible for bringing Ryzen to market-"you don't have any idea how many hours I and my team have spent on this," Anderson said-it's Papermaster who has to think of the future. When asked how long Zen would last, compared to Intel's two-year tick-tock cadence, Papermaster confirmed the four-year lifespan and tapped the table in front of him: "We're not going tick-tock," he said. "Zen is going to be tock, tock, tock."
Difference, though, is that Intel hasn't done a from-scratch design in a long, long time; they're basically trying to squeeze blood from the P6/Conroe turnip. Zen is new, so there's almost certainly more low-hanging fruit they can address to improve things.
If they entirely focused on clockspeed, they could potentially reach 5Ghz (my guess is 4.5-4.6ghz for the record), still 10-12% better single core ain't bad.
Thanks for posting this, it seems to dispel the oft-repeated claim that 12nm Zen is nothing more than an optical shrink.Originally Posted by WannaBeOCer
I am expecting performance gains. AMD said it themselves regarding Zen.
https://www.pcworld.com/article/3155129/components-processors/amd-says-its-zen-cpu-architecture-is-expected-to-last-four-years.html
Quote:
While Anderson's responsible for bringing Ryzen to market-"you don't have any idea how many hours I and my team have spent on this," Anderson said-it's Papermaster who has to think of the future. When asked how long Zen would last, compared to Intel's two-year tick-tock cadence, Papermaster confirmed the four-year lifespan and tapped the table in front of him: "We're not going tick-tock," he said. "Zen is going to be tock, tock, tock."
This is a brand-new design that got relatively rushed out the door. Engineers have said there's a lot of low-hanging fruit. Just look at Bulldozer to Piledriver; some small changes to the caches, front-end, and load/store units and you got ~15% IPC uplift in exactly one year on what was technically labelled by AMD as a new stepping of the same die. I think you are far more likely to see a larger IPC improvement than quoted than to see 5GHz. GF only estimates a 10% performance improvement moving to 12nm, which would translate to 4.4GHz on the flagship, and getting the rest of that via design changes is not easy to do without sacrificing IPC.Originally Posted by Buris
If they entirely focused on clockspeed, they could potentially reach 5Ghz (my guess is 4.5-4.6ghz for the record), still 10-12% better single core ain't bad.
5Ghz, BAM 20% uplift achieved on per-core performance, along with higher frequency RAM etc.
IPC probably will see an extremely low gain (something like 1-3%), as this is more of a quick Zen improvement and not a massive rework of the architecture. I imagine 7nm Zen 2 will be the massive improvement on IPC/Power efficiency.